You may have noticed a sharp increase in problems and issues in Windows recently – following the rise of the “AI” hype cycle, entirely coincidentally, I’m sure – and it seems Microsoft is finally starting to acknowledge just how bad Windows has become.
On the positive side though, following all that backlash, Microsoft acknowledged Windows has issues, and as if on cue, the company in a new support article has admitted that there are problems on almost every major Windows 11 core feature. The issues are related to XAML and this impacts all the Shell components like the Start Menu, Taskbar, Explorer, and Windows Settings.
↫ Sayan Sen at Neowin
It’s wild how many core components like this have apparently been broken due to these problems since July of this year. This means countless Windows users have been experiencing weird issues on a daily basis in multiple components for four months now, which is absolutely wild. On top of all the more structural problems in Windows, I wonder how people can get anything done at all – only a few days ago, I had to manually clean out the Installer folder in the Windows folder on my wife’s gaming PC, because for some inexplicable reason, Windows decided to permanently store 18GB’s worth (!) of past Adobe Acrobat updates and installers in there.
It’s impossible to reliably say that Microsoft’s incessant focus on crypto NFTs “AI” lies at the root of all of these problems, but if 30% of “new” code in Microsoft is indeed regurgitated by “AI”, it’s hard not to conclude as such.

Time to replace some CEO and CTO by AI. What can go more wrong ?
Given that current CEOs and CTOs are basically Excel-file-numbers-up driven, AI would probably be an improvement. At least AI tends to agree that the user is right.
With the AI integration in Excel, Microsoft management would become AI slop.
It’s already slop, just slop with huge inertia
Kochise,
I wish, but at those levels “human connection” becomes much more important.
CEOs are for the show, they would go to a local restaurant and give “humble” photo shoots, they would meet heads of states to get favorable tax breaks, and they would meet other rich people in a ball game to get billions in investments.
AI cannot do that.
CTO? Maybe. I’m not sure what they would do from now on.
> because for some inexplicable reason, Windows decided to permanently store 18GB’s worth (!) of past Adobe Acrobat updates and installers in there.
To be fair, that specific problem could just as well have been caused by adobe.
Mote,
I would insist it is most likely Adobe.
(As much as I like their products, their feature set has no competition, the ecosystem around is “questionable” for lack of a better word)
I’ve been a lifelong Windows user. A few weeks ago I got a new ThinkPad and it came with Windows 11 preinstalled. The first thing I did was to uninstall the bloatware, and left just the Lenovo software for keeping the BIOS up to date. I then proceeded to partition the SSD, and installed the latest Pop!_OS beta as my main OS. I will not be using Windows in the future if I can help it.
@FriendBesto
We were almost all lifelong Windows users until we were not. Welcome aboard.
– Hello my name is Kochise, I’m a Windows user.
– Welcome Kochise…
I was looking around at new laptops and I think my next one might be a Chromebook. I have owned one before, my error then was in getting an ARM-based one because it greatly limited the available software. The machines are known to work well as Linux ultrabooks if you shop around for compatibility, and even the stock OS will now let you run a VM to access Linux apps.
Windows, on the other hand, has been relegated to the machine that runs gaming and some audio apps. I can see myself migrating entirely off it in a short time, with a little bit of pain.
I’m not entirely sure if Microsoft wants to be a consumer oriented company anymore.
More and more it has this IBM vibe on it, going deeper to a B2B only route, and treating final individual consumers more like a liability or legacy that they don’t know how to get rid of them than actual customers.
I’ve said this in the past, I feel like Microsoft has a specific goal for consumer PCs: 100% cloud computing. At some point in the near future when you buy a new PC, the version of Windows it comes with will be a boot stub. It will boot into a “cloud” provided full operating system, with no local user-accessible storage. You will always be running the latest possible release from Microsoft, with no way to roll back or decide which “version” you want to run. The vast majority of computer users won’t notice or care; they might see degraded performance if their Internet connection drops or is unstable, but as long as the machine is powered on the OS will stay resident in RAM and they will go about their business just like they did when Windows was something you booted locally. Of course this will be subscription based; if you don’t pay up, you lose access to your data and your computer becomes useless.
And honestly, I find the concept intriguing and practical, to a certain point. The obvious downside is, of course, the fact that you truly don’t own your computer anymore. Without Microsoft’s online OS, it’s a brick. There’s no local storage and no way to add it, there’s no way to load a different OS, the boot stub area is just large enough to hold Microsoft’s code and is read-only anyway, with keys owned by them not you.
Google has already proven that this model works in the consumer space with Chromebooks, but they don’t have a large enough market share to make it universal. Microsoft does, and they have every reason to move forward with this paradigm and little reason to hold back. This is the future they have decided you’ll live with, it’s just a matter of time before it’s realized.
ChromeOS is more local than people give it credit for. Generally, having an OS that it video streamed (like Google Stadia) sounds like a good idea, but impossible outside of corporate buildings with fast LANs. Same for OSes that boot from an NFS drive (like Sun’s “diskless workstations” of old): it works in corporate buildings with fast reliable LANs, but will never work in something like a laptop that has to boot inside an airplane with no in-flight WiFi or will spend considerable time connected to dodgy WiFi hotspots in cafeterias.
What we may see is a local OS that is tied to an online account just for subscription billing and spying purposes, much like Adobe’s Creative “Cloud”: a local app where the “cloud” component is there just to enable subscription billing and to enable Adobe to train their AI on your works (I know they backtracked on this, but the fact they were about to do it means they can do it). Windows 11 forcing a Microsoft Account is a precursor to this. Yes, you can bypass it for now and create a local account, but Windows 12 may remove local accounts entirely.
@Morgan
I agree with you. It is on old idea of course back to mainframes and X terminals. You could say it Sun’s vision of “The Network is the Computer” which Sun coined in 1984 (same year as the first Macintosh) and Cloudflare recently trademarked. What is old is new again.
The big difference of course is who owns the resources you command. What is new is that everything is a subscription.
I think by the time they have this ready Linux will be looking/working like Mac OSX. I really don’t see this idea being popular with anyone. Microsoft is now in a race against AI to remain relevant. Remember, while AI coding tools help Microsoft they are also Microsoft’s weakness. Anyone can now replicate Microsoft’s software leveraging AI tools to make functional equivalents to Windows Software. We are already starting to see an acceleration in open source projects. I needed to make my Linux desktop viewable inside of VR, I was able to write an application in two days using ChatGPT. There’s no way I would have had the skills/knowledge to develop that application without months of reading/learning about poses/graphics/texture copy methods. This is the biggest threat to Microsoft’s dominance and they are cheerleading it. Being able to add major useful features to open source software without having to wait for someone else to be interested in making it is a fundamental shift we’ve never seen in Open Source Software.
“I think by the time they have this ready Linux will be looking/working like Mac OSX. I really don’t see this idea being popular with anyone.”
The thing is though, it doesn’t have to become popular with consumers to become the norm. This has already been proven with the abominable forced updates that started in Windows 10. Users hated them, but most didn’t care enough to actually abandon Windows altogether, and that hated functionality has now become the norm. The same loathing followed by begrudging acceptance now seems to be happening with many aspects of Windows 11. And for the small minority of us that actually do want to run an alternative OS, there’s nothing to guarantee that those “cloud-boot” devices won’t be every bit as locked down as the average smartphone. So personally I am rather concerned and very pessimistic about the direction that Big Tech will probably take the market, with no regard whatsoever for us non-conforming weirdos.
rahim123,
I agree this is a real problem. It’s all well and good that there are some people who actively fight the corporate control agenda, but if our numbers become so marginalized that corporations can ignore us as a group, it may start to close the doors for all but the most determined in the future.
Just recently google announced their plan to take away owner sideloading rights on android. We dodged a bullet when they backtracked over all the bad press (thank you Thom) , but the detrimental impacts would have been far reaching. Microsoft’s secure boot policy could technically have easily killed linux on commodity x86 PCs, fortunately there was enough backlash and manufacturers let owners disable it. But those terms have since expired and microsoft are still in possession of a secure boot nuclear option (assuming they can plot a course through antitrust waters and nobody stopped them).
We need to stay vigilant! If we keep loosing ground, even a little bit at a time, our freedoms can die through a thousand cuts.
Morgan,
The future you describe reinvents IBM mainframes, which themselves were rented out under contracts. Back then this model took hold because computers were inaccessible even for businesses. As unix & windows PC became affordable on desktop and as servers, those quickly replaced the mainframe model for the vast majority of the world. I don’t think users are demanding to go back, but it is clear that corporate interests want consumers to be dependent on their centralized services.
While I’m not strictly against the remote technology, especially if the OS and standards are vendor agnostic and allow self-hosting, however the problem is obvious: corporations aren’t engineering the technology in a way that empowers owners to have control and freedom of choice. Instead we get more locked down. Google are guilty of this, apple are guilty of this, and although they’ve stumbled, MS want to be guilty of this too. A world where corporations keep full control of everyone’s hardware and data is incredibly dystopian.
@CapEnt
Microsoft makes all their money from Azure and Office. That tells you all you need to know about where their interests lie. Windows is a platform to run Office and a kernel to run games on. Sometime soon, they will switch to thinking of Office as running entirely on Azure and will stop caring OS talks to your keyboard.
For the rest of the herd, it almost makes sense for them to harvest data and ad revenue.
In all fairness, Microsoft didn’t get into cryptocurrency, despite cryptocurrency being mainstream for some years now.
They also didn’t get into NFTs, presumably because it was obvious even to Microsoft execs that NFTs were a “greater fool” scam. Samsung did get into NFTs by allowing you to buy and view NFTs on your Samsung TV (because that totally adds value compared to loading up a bunch of JPEGs and viewing them in a slideshow), so you may be thinking of that, but Microsoft didn’t get into NFTs (not on client-facing products anyway, Azure has an NFT transaction accelerator as a minor feature).
But Microsoft going all-in on “AI” (LLMs) is of course real, Windows 11 is being progressively crapified with “agentic” stuff that nobody wants at the OS level and the rest of Windows 11 is crumbling or even made worse (whether it’s due to LLM-written code, we’ll never know). I plan to stay on Windows 10 ESU for as long as possible.
(note: no, I’m not arguing with myself, I didn’t write the entirety of the article)
kurkosdr,
There is a very valid reason they want to catch the AI (or rather “foundation model”) bandwagon.
They tried this with “assistants” but failed (who remembers Cortana)? But the entire assistant ecosytem failed as it was living in the “uncanny valley”. Sounds as if they are intelligent, but they were nothing more than very simple pattern recognition engines.
Actual voice interactions is the next frontier of user interactions. We had the punch cards, we had the terminal, we had mouse based UIs, we had touch screens, now the computers will “speak” (and think).
It is definitely not ready today, but like everything else in Star Trek, they are on the way to becoming the norm.
sukru,
Ironically, all of the “inventions” to come out of star trek, some breaking the known laws of physics, computing is one where I don’t believe the show was thinking far enough ahead. The ship computer in TOS and even TNG were very basic even by today’s standards. Of course Data, the android, might somewhat make up for that, but in the star trek universe Data was exceptional and computers they were using were quite dumb even by today’s standards.
Of course for narrative purposes it might be better to avoid everything becoming automated by computer – even if that’s one of the more obvious outcomes of AI. Also, why bother with automation when the red shirts are expendable 🙂
Alfman,
They innovations were more on the “user experience” part.
They could not get good modular UIs, so they build “touch screens” (LCARS). They could not get email, so they built “tablets”, they could not get proper command systems, so they used “voice control” (“computer, tea, earl grey, hot”)
This is similar to how they invented “transporters”. It was difficult for them to do the CGI for shuttles, so they teleported people with much simpler effects.
If you think about these, they are all natural evolutions of user – computer interactions. That is why voice, or rather “language” is the next frontier that is being built today.
(Remember the scene where Scotty was trying to convene with a mouse on an ancient machintosh)?
sukru,
Yeah, I understand the show had limited budget for props. As a scifi show they deserve some creative space and I can’t be too hard on them even though some of the technology depicting the future is already antiquated.
It was a funny seen. It was a decent joke, but more than that it cracks me up that Scotty would even know how to program some archaic computer like that. It’s rare for media to treat technology realistically.
Still, it could be a lot worse…
“NCIS 2 IDIOTS 1 KEYBOARD”
https://www.youtube.com/watch?v=u8qgehH3kEQ
Alfman,
I could easily envision Scotty being a ned that is interested in “retro” stuff. After all don’t many of us collect and revive old antique devices (though not for daily use, they are extremely inefficient and clunky)?
I don’t mind them experimenting with “agentic” stuff as long as 1) it’s easy to turn that stuff off (basically a settings page titled “Agentic stuff” with toggles for all the agentic/AI stuff would be great) and 2) they don’t let the rest of the OS degrade by letting interns or (even worse) LLM bots write buggy code for it
The case of modern Microsoft will be studied in future business schools in the same manner as the case of GM in the 80s and 90s, where they experimented with all kinds of futuristic ideas in the chase for their next big thing (from highly automated factories to cylinder deactivation to touchscreen controls to fully electric cars, all of which were indeed the future), but the quality of the product suffered due to silly whoopsies, because all the R&D money was diverted to the next big thing (you see, GM’s accountants would let investment in both).
I mean, Windows is already seeing its market share slowly be eroded by MacOS (especially among people who want more than a dirt-cheap browsing and e-mail machine), and Valve is preparing to launch a Steam Machine that will be a competitively-priced pre-built. Of all the times to make Windows worse, this is the worst time.
kurkosdr,
As you have implied, unfortunately those things don’t stay silo’ed.
Either the code will creep into the main codebase, or they will take the time away from other crucial functions. Either way, “non AI” interface will suffer.
The most optimal way of doing this would be integrating with existing accessibility frameworks. This would have many benefits:
1 – It won’t break anything that is currently working
2 – It would even benefit accessibility if they discover gaps in the user experience. After all, they will have to prioritize them now.
2.5 – Apps will be updated to fix accesibility.
3 – It will be “agnostic”. Another company like OpenAI, Google, or even open source ones like ollama, or llama.cpp could take over the “agent” functions, if there is a great API to connect humans to the computer. (Or humans’ agents)
I still believe agents are the future. They would be part of past designs of the Internet. We were dreaming of a future that never came where our agents would do work on our behalf, on a peer-to-peer basis.
But I’m not holding my breath. I don’t think MS would be principled enough to do it this way. Especially when they have more or less completely de-prioritized windows (was it less than 10% of their revenue in their last report)?
“Microsoft admits almost all major Windows 11 core features are broken”
I vote that almost all major Windows 11 core features are not broken.